Suppose you develop a new weight lose drug and want to demonstrate its
effectiveness. You put an advertisement in the local newspaper asking
for volunteers who want to try the drug.
A bunch of people respond to your ad. You invite them all to your office, and tell them about how your wonderful new drug will help them lose weight. You then weigh all of the volunteers (to establish a baseline measurement), and then send them home with instructions on how to take the drug.
Two weeks later, you invite them back to your office for another weighing. You are delighted to discover that the average person lost 5 kg.
(Besides being illegal), what are some problems with this research design? And what would you do differently?
Key Idea: because the people in C are, on average, the same as the people in T, we have removed compositional (selection) effects through randomization. We can thus estimate the causal effect of treatment.
Suppose your friend sees Enos’ results and says:
“Nice experiment, but he’s just documenting a temporary reaction to the unexpected appearance of Latinos in all-white suburbs. Over time, however, people are going to become more comfortable with diversity. For example, cities have historically been magnets for immigration, and the people living there seem to have no problem with diversity.”
What do you think about this argument? What about the “evidence” your friend cites?
Think-pair share: How would you modify Enos’ experiment to test the hypothesis that prolonged contact can moderate the negative influence of exposure to immigration?